John A. Hossack, PhD

John A. Hossack, PhD

[Dynamic Data - Faculty Directory ]

Ultrasound imaging of prostate cancer

Ultrasound Prostate Imaging - 3D Anatomic Assessment and 'Synthetic Digital Rectal Examination'

This research in the Hossack laboratory is directed towards the earlier, more quantifiable and more repeatable diagnosis of prostate cancer. This is motivated by the inadequacy of current methods reliant on the PSA blood test and the Digital Rectal Exam (DRE). The "I-Beam" approach that they are pioneering enables freehand 3D ultrasound imaging capability using two 'Tracking' ultrasound imaging arrays orthogonally mounted on the two ends of the conventional (centrally placed) ultrasound 'Imaging' array. When viewed from directly above the array, the imaging planes of the central and two tracking arrays form an 'I' shape - hence 'I-Beam'. This approach is compatible with current ultrasound biplane transrectal probes and is very low cost to implement. This approach has been shown to yield a 2s linear dimensional accuracy of 4.6%.

The transducer also has a water injection port. As pressure from the latex sheath over the transducer due to the water injection is applied to the tissue (or phantom) and minor deformation is generated in both the lesion and the surrounding tissue. Deformations of lesion and tissue can be distinguished because the strains in each are different due to their different moduli of elasticity. To determine the deformation, the displacement coefficients between two image frames were calculated on the raw ultrasound data using cross-correlation. The correlation lag index corresponding to the greatest cross-correlation coefficient, i.e., the best match was calculated for each search window, in this case with a size of four RF cycles in axial dimension.  Subsample precision is obtained by using quadratic interpolation around the cross-correlation best match. Repeating this process throughout the raw ultrasound data of the image frame created a displacement map between two frames. Finally, a strain image is obtained by spatial differentiation, in the range dimension, of the displacement map.